BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//wp-events-plugin.com//7.4.0.1//EN
TZID:America/New_York
X-WR-TIMEZONE:America/New_York
BEGIN:VEVENT
UID:631@fds.yale.edu
DTSTART;TZID=America/New_York:20240918T113000
DTEND;TZID=America/New_York:20240918T130000
DTSTAMP:20250916T142138Z
URL:https://fds.yale.edu/events/fds-colloquium-josh-alman-columbia/
SUMMARY:FDS Colloquium: Josh Alman (Columbia)\, "Algorithms and Hardness fo
 r Kernel Density Estimation and Attention"
DESCRIPTION:Webcast Link (via Zoom - starts at 12:00): https://yale.zoom.us
 /s/7859884026\n\n\n\n"Algorithms and Hardness for Kernel Density Estimatio
 n and Attention"\n\n\n\nAbstract:  This talk will focus on two related com
 putational problems. The first is Kernel Density Estimation\, a statistica
 l task which has diverse applications from machine learning to computatio
 nal physics. The second is Attention\, the task at the core of Transformer
  and other Large Language Models. \n\n\n\nBoth of these problems have str
 aightforward quadratic-time algorithms\, and I'll discuss a recent line of
  work investigating when faster algorithms are possible. I'll survey two a
 lgorithmic techniques which lead to almost linear-time algorithms in diffe
 rent parameter regimes: the Fast Multipole Method\, and the polynomial met
 hod. I'll then overview our new fine-grained hardness results\, which prov
 e that these techniques are essentially optimal in the parameter regimes w
 here they apply\, and highlight the situations where improvements may be p
 ossible.\n\n\n\nBased on joint work with Amol Aggarwal (in CCC 2022)\, Zha
 o Song (in NeurIPS 2023)\, and Yunfeng Guan (in CCC 2024).\n\n\n\nBio: Jos
 h Alman is an Assistant Professor of Computer Science at Columbia Univers
 ity. He is a theoretical computer scientist who works on algorithm design 
 and complexity theory\, focusing especially on algebraic tools for speedin
 g up algorithms.\n\n\n\nWebsite: joshalman.com/\n\n\n\n\n\n\n\n\n
CATEGORIES:FDS Events,Colloquium,Seminar Series
END:VEVENT
BEGIN:VTIMEZONE
TZID:America/New_York
X-LIC-LOCATION:America/New_York
BEGIN:DAYLIGHT
DTSTART:20240310T030000
TZOFFSETFROM:-0500
TZOFFSETTO:-0400
TZNAME:EDT
END:DAYLIGHT
END:VTIMEZONE
END:VCALENDAR